Replacing the dev-test-deploy-ops rut with a collaborative approach to delivering business app services

DevOps for Business Application Services

Subscribe to DevOps for Business Application Services: eMailAlertsEmail Alerts newslettersWeekly Newsletters
Get DevOps for Business Application Services: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn

DevOps Authors: Zakia Bouachraoui, Pat Romanski, Yeshim Deniz, Liz McMillan, Elizabeth White

Blog Feed Post

What is a Hyper-Converged Network (and why do you need one)?

Certainly by now everyone in IT has heard of Hyper-Converged Infrastructure (HCI). In the case that you were abducted and held by aliens for the past two years, let me explain. HCI is loosely defined as the close coupling of virtualized compute and virtualized scale-out storage, typically mated to a server platform to create an easy-to-use appliance. So, what does this have to do with networking? Despite the close coupling of storage and compute, not all those needs fit into a single HCI appliance (nor would a scale-out approach be particularly interesting if it were limited to a single device). That means the appliances part of the HCI need to be able to actually talk to each other. How? Over a network.

Sure, most people already have a network, so why isn’t that enough? The simple answer is that HCI appliances have unique needs that existing legacy networks were never designed to handle. And those networks have their own operational silo, keeping the benefits of convergence (reduce silo’s, more agile IT, and lower operating costs) from fully impacting the business. We’ll explore these a bit more later in this post and subsequent posts.

The networking we’re talking about here is different from the virtual network services that can run over the physical network. That physical network, the one that provides the “plumbing” for the HCI appliances to communicate over, is what we call the network fabric. That fabric increasingly needs to support specific services that easily extend and augment the ease-of-use, performance, and security promises of HCI. This results in the need for what is referred to as a Hyper-Converged Network (HCN).

So what is a HCN?

HCN is purpose-built network fabric to support the unique needs of HCI clusters. It’s an API-driven network, rather than protocol driven, and better than that, it’s already fully integrated into your HCI management system, so it automatically:

  • Discovers VMs
  • Creates VLANs
  • Automatically creates an isolated storage network within the network fabric to ensure optimal performance as the HCI appliances exchange data, meta-data and other critical storage traffic
  • Discovers security groupings and policies implemented at the hypervisor and creates their equivalent ACLs in the network
  • Provides visualization of the network in the context of your HCI and easily points the operator to potential trouble areas

HCN is a physical Ethernet/IP-based network fabric that connects HCI appliances within a single rack, across multiple racks, across pods, and across data centers and converges the “LAN” and the “SAN” in a single fabric. Romantic, right? It’s a scale-out network that can simply add ports and network bandwidth as switches are added, without needing separate scale-up tiers, like “spines” or “aggregation” switches. HCN is practically fool proof – it’s easy to cable, and switches self-discover each other to form fabric without user intervention, and with no configuration protocols.

Why do I need one?

Great question – and here are three great answers:

  • It’s integrated. HCI has increasingly become the way to build private clouds in the Enterprise. The promise of self-service IT needs to break down the networking silo barrier to be successful, and this can happen when you have a network that is directly integrated with your HCI deployment.
  • Life’s too short for protocols. Endless tweaking of timers and other arcane settings is like pushing a noodle, you never really get what you want. HCNs are fast, repeatable, less prone to errors, and actually gets the desired result. Networks, like all IT infrastructure, need to be API-driven to be automated.
  • Because no one is going back to the days of SANs. HCI removes the need for Storage Area Networks. But SANs ensured that storage system traffic was reliable and performant, and kept that traffic isolated from the LAN. It also did things behind the scenes like ensure critical meta-data traffic was never lost and delivered with the lowest latency. HCNs let you create a converged network fabric that is partitioned for the storage traffic, so you get the benefit of convergence with the performance of dedicated LANs and SANs.

And if that’s not enough, we’ll leave you with one bonus reason:

  • It costs you the same or less than doing it the hard way. That’s right, a Plexxi HCN can not only save you considerable time and frustration, but it also costs the same or less than traditional networking approaches. That does not even factor in the savings you would get by better leveraging the available fabric capacity (so you get more for less and upgrade less frequently) or the cost of having separate dedicated networks for compute and storage!



The post What is a Hyper-Converged Network (and why do you need one)? appeared first on Plexxi.

Read the original blog entry...

More Stories By Mat Mathews

Visionary solutions are built by visionary leaders. Plexxi co-founder and Vice President of Product Management Mat Mathews has spent 20 years in the networking industry observing, experimenting and ultimately honing his technology vision. The resulting product — a combination of traditional networking, software-defined networking and photonic switching — represents the best of Mat's career experiences. Prior to Plexxi, Mat held VP of Product Management roles at Arbor Networks and Crossbeam Systems. Mat began his career as a software engineer for Wellfleet Communications, building high speed Frame Relay Switches for the carrier market. Mat holds a Bachelors of Science in Computer Systems Engineering from the University of Massachusetts at Amherst.